Fuzzing is an industry-standard software testing technique that uncovers bugs in a target program by executing it with mutated inputs. Over the lifecycle of a fuzzing campaign, the fuzzer accumulates inputs inducing new and interesting target behaviors, drawing from these inputs for further mutation and generation of new inputs. This rapidly results in a large pool of inputs to select from, making it challenging to quickly determine the "most promising" input for mutation. Reinforcement learning (RL) provides a natural solution to this seed scheduling problem---a fuzzer can dynamically adapt its selection strategy by learning from past results. However, existing RL approaches are (a) computationally expensive (reducing fuzzer throughput), and/or (b) require hyperparameter tuning (reducing generality across targets and input types). To this end, we propose T-Scheduler, a seed scheduler built upon multi-armed bandit theory to automatically adapt to the target. Notably, our formulation does not require the user to select or tune hyperparameters and is therefore easily generalizable across different targets. We evaluate T-Scheduler over 35 CPU-yr fuzzing effort, comparing it to 11 state-of-the-art schedulers. Our results show that T-Scheduler improves on these 11 schedulers on both bug-finding and coverage-expansion abilities.

Modern processors have multiple pipelined functional units and can issue more than one instruction per clock cycle. This places a burden on the compiler to schedule the instructions to take maximum advantage of the underlying hardware. Superblocks - a straight-line sequence of code with a single entry point and multiple possible exit points - are a commonly used scheduling region within compilers. Superblock scheduling is NP-complete, and is done suboptimally in production compilers using a greedy algorithm coupled with a heuristic. The heuristic is usually handcrafted, a potentially time-consuming process. In this paper, we show that supervised machine learning techniques can be used to semiautomate the construction of heuristics for superblock scheduling. In our approach, labeled training data were produced using an optimal superblock scheduler. A decision tree learning algorithm was then used to induce a heuristic from the training data. The automatically constructed decision tree heuristic was compared against the best previously proposed, handcrafted heuristics for superblock scheduling on the SPEC 2000 and MediaBench benchmark suites. On these benchmark suites, the decision tree heuristic reduced the number of superblocks that were not optimally scheduled by up to 38 percent, and led to improved performance on some architectural models and competitive performance on others.

The National Ice Center (NIC) routinely provides weekly global sea ice analyses, daily analyzed and forecasted ice edge products, and tailored high-resolution satellite imagery in support of various U.S. Government agency requirements. Heavy reliance is placed on visible and infrared imagery from the civilian and defense meteorological satellites; however, in areas of persistent cloud cover the NIC takes advantage of passive microwave satellite data to fill these gaps. Currently the NIC analysts use the NIC Hybrid sea ice algorithm in the Northern Hemisphere and the Bootstrap sea ice algorithm in the Southern Hemisphere to complete their products. This particular choice of algorithms was initially based on the personal preference of individual analysts, and was then perpetuated through ongoing training programs. No rigorous assessment of the merit of each algorithm was undertaken. In July 2001 the NIC began an ongoing evaluation of the NASA Team, Bootstrap, NIC Hybrid, and NASA Team II sea ice algorithms. The evaluation was performed by comparing available, time-coincident, visible and infrared satellite imagery with the SSM/I products. To date, 13 cases have been collected and analyzed. The NASA Team II provides the consistently best estimate of the ice edge and the 90% concentration boundary. This paper discusses the study undertaken by the NIC and presents select case studies to illustrate the performance characteristics of the various algorithms.